Essays on Decision Making with Information and Action Externalities
In this thesis, I investigate the behavior of rational individuals making decisions involving externalities and explain why seemingly irrational behavior can be the result of the decisions of rational individuals. In particular, I consider two kinds of externalities: an information externality and an action externality. In Chapter two, I characterize the equilibria of a dynamic game with an information externality and an action externality where players with private information make decisions in a sequential but exogenous order. I investigate the effects of different job evaluation criteria on the players' behavior in a two-stage model. I find that if the evaluation is based on a relative measure instead of an absolute measure, herding may occur much earlier, resulting in less efficient information aggregation and heavier loss in social welfare. In Chapter three, I characterize the asymmetric pure strategy equilibria of a dynamic game with a pure information externality where players choose whether to invest as well as when to invest. I find that any asymmetric pure strategy equilibrium is more efficient than the symmetric mixed strategy equilibrium which is the focus of the existing literature. I also find that when players become very patient, the least efficient asymmetric pure strategy equilibrium resembles those equilibria in an exogenous timing model where players make decisions one by one in an exogenous order. In Chapter four, I analyse the voters' decision to vote in large elections. I reexamine the paradox of voting by incorporating modern psychological and sociological findings and model elections as something similar to spectator sports events. I find that in large elections, the voting rate is significantly different from zero, and thus voting in large elections is justified. Moreover, the voting rate is higher for the candidate who has more supporters. Using the polls before the election as a proxy for the number of supporters a candidate has, I show that the data from the 2004 US Presidential Election support this theory.